Auditory experience shapes how listeners exploit perceptual equivalency classes in speech and cope with noise degradation: Evidence from “elliptical speech”

Purpose: Speech perception requires that listeners classify sensory information into smaller groupings while also coping with noise that often corrupts the speech signal. The strength of categorization and speech-in-noise (SIN) abilities show stark individual differences. Some listeners perceive speech sounds in a gradient fashion, while others categorize in a discrete/binary manner, favoring fine acoustic details vs. a more abstract phonetic code, respectively. Prior work suggests SIN processing is (i) related to more gradient phonetic perception and (ii) varies with musical training. Method: To further probe relations between perceptual gradiency and noise-degraded listening, we measured phoneme categorization, SIN recognition (QuickSIN), and sentence recognition in listeners with varying musical backgrounds. Categorization was measured for vowels and stops using standard labeling tasks. Speech recognition and discrimination were assessed using "elliptical speech" sentences that use featural substitutions which renders them meaningless under clean conditions but surprisingly improves their recognition under noise degradation. We hypothesized listeners who use broader perceptual equivalency classes in hearing elliptical speech would show better SIN perception, indicative of a more gradient listening strategy. Results: Listeners perceived elliptical sentences as sounding different than their intact counterparts in the clear but as the same under noise degradation. But this elliptical benefit varied with music background. Nonmusicians showed larger susceptibility and noise-related benefit of ellipses than musicians, consistent with the notion they used broader phonetic categories (i.e., more gradient listening). Elliptical speech perception was also associated with QuickSIN performance in both groups but in opposite ways. Conclusions: Use of broader categories was related to better SIN processing in nonmusicians but poorer SIN processing in musicians. Findings suggest listeners can use broader perceptual equivalence classes to deal with degraded listening situations but this depends critically on their auditory demographics. Nonmusicians might use broader phonetic categories to aid SIN perception while musicians might use narrower categories or otherwise similar speech contexts.